[OpenAFS] user home directory replication
Richard Brittain
Richard.Brittain@dartmouth.edu
Thu, 15 Jul 2010 12:13:03 -0400 (EDT)
This message is in MIME format. The first part should be readable text,
while the remaining parts are likely unreadable without MIME-aware tools.
---1434362608-623761406-1279210384=:16797
Content-Type: TEXT/PLAIN; charset=UTF-8; format=flowed
Content-Transfer-Encoding: quoted-printable
X-MIME-Autoconverted: from 8bit to quoted-printable by mailhub1.dartmouth.edu id o6FGD3Y6016628
On Wed, 14 Jul 2010, Jonathan Nilsson wrote:
> I would like to replicate home directories (and other AFS volumes that =
are primarily
> accessed read/write) for the purpose of faster disaster recovery in cer=
tain common cases,
> such as local hardware failure on an AFS File Server.=C2=A0 I am planni=
ng to do this by always
> mounting these volumes with the "-rw" flag, then simply adding a replic=
ation site, and
> running a script that will "vos release" all these volumes every so oft=
en.
We don't do this for homes, but we do it for auxiliary data volumes (100G=
B=20
and up), which we don't usually send to our tape backup system. The=20
normal user access path is via explicit RW mount, and we mount the=20
RO replicate, as well as yesterday's .backup, under well defined paths so=
=20
that the users can dig into them if they feel the need.
What I wish we had is a -dryrun and -showfiles flag for vos release, so I=
=20
can see what files would be copied and how big they are, before I do the=20
copy. I auto-release several TB of data daily and with my normal user=20
activity, it only takes 15-30 mins for the incremental releases, but=20
occasionally much longer. I wish we had those options available to=20
incremental 'vos dump' too.
We have separate management tools (currently clunky) to make sure that th=
e=20
replicates are on a different server, in a different machine room, but=20
that side of things is far from perfect yet. We don't have the server=20
space to allocate a RO partition to match each RW partition, so stuff is=20
mixed up.
I'm still trying to figure out how to make more copies available using=20
shadow and clone volumes, but for a single replicate, the mature RO=20
replicate and 'vos release' works great.
Richard
--=20
Richard Brittain, Research Computing Group,
Kiewit Computing Services, 6224 Baker/Berry Library
Dartmouth College, Hanover NH 03755
Richard.Brittain@dartmouth.edu 6-2085
---1434362608-623761406-1279210384=:16797--